Text-to-Image Retrieval on RSICD (test)
15.59R@1GeoRSCLIP-FT
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| GeoRSCLIP-FTParadigm=GVLM + FT, Tuned on=RS5M + RET-22023.06 | 15.59 | 41.19 | 57.99 | 38.87 | — | |
| GeoRSCLIP-FTParadigm=GVLM + FT, Tuned on=RS5M + RSICD2023.06 | 15.26 | 40.46 | 57.79 | 38 | — | |
| ELC0Backbone=ViT-B/322026.07 | 14.03 | 37.37 | 54.38 | — | 213.47 | |
| ELCBackbone=ViT-B/322026.07 | 14.03 | 38.01 | 54.71 | — | 215.2 | |
| RemoteCLIPParadigm=GVLM + FT, Tuned on=RET-3 + DET-10 + SEG-42023.06 | 13.71 | 37.11 | 54.25 | 35.26 | — | |
| RemoteCLIPBackbone=ViT-B/322026.07 | 13.71 | 37.11 | 54.25 | — | 211.57 | |
| FGVLABackbone=ViT-B/322026.07 | 12.72 | 36.61 | 53.76 | — | 204.74 | |
| PAUBackbone=ViT-B/322026.07 | 12.39 | 35 | 53.47 | — | 198.39 | |
| AIRBackbone=ViT-B/322026.07 | 12.19 | 37.07 | 53.94 | — | 199.46 | |
| InfoNCEBackbone=ViT-B/322026.07 | 11.89 | 35.31 | 53.14 | — | 196.49 | |
| GLGSABackbone=ViT-B/322026.07 | 11.75 | 34.07 | 50.12 | — | 192.4 | |
| PE-RSITRParadigm=GVLM + FT, Tuned on=RSICD2023.06 | 11.63 | 33.92 | 50.73 | 31.12 | — | |
| PE-RSITRBackbone=ViT-B/322026.07 | 11.63 | 33.92 | 50.73 | — | 186.7 | |
| GeoAlignCLIPBackbone=ViT-L/142026.03 | 11.4 | 31.29 | 44.34 | — | — | |
| FGCLIPBackbone=ViT-L/14-3362026.03 | 11.18 | 29.83 | 43.31 | — | — | |
| VSE∞Backbone=ViT-B/322026.07 | 11.18 | 32.83 | 47.7 | — | 193.08 | |
| VSE++Backbone=ViT-B/322026.07 | 10.74 | 30.41 | 45.56 | — | 185.35 | |
| PolyBackbone=ViT-B/322026.07 | 10.48 | 31.75 | 46.18 | — | 181.18 | |
| GeoRSCLIPBackbone=ViT-L/14-3362026.03 | 10.36 | 28.44 | 42.09 | — | — | |
| LRSCLIPBackbone=ViT-L/142026.03 | 10.01 | 28.6 | 43.73 | — | — | |
| GeoRSCLIPBackbone=ViT-L/142026.03 | 9.92 | 28.01 | 42.09 | — | — | |
| CUPBackbone=ViT-B/322026.07 | 9.86 | 31.6 | 47.15 | — | 182.02 | |
| GeoAlignCLIPBackbone=ViT-B/162026.03 | 9.55 | 28.07 | 41.92 | — | — | |
| GeoRSCLIPParadigm=GVLM + FT, Tuned on=RS5M2023.06 | 9.52 | 27.37 | 40.99 | 26.18 | — | |
| LRSCLIPBackbone=ViT-B/162026.03 | 9.15 | 27.43 | 39.65 | — | — | |
| MultilanguageParadigm=Supervised, Tuned on=RSICD2023.06 | 9.14 | 28.96 | 44.59 | 27.42 | — | |
| SIRSBackbone=ViT-B/322026.07 | 8.74 | 30.38 | 49.5 | — | 166.31 | |
| MTGFEParadigm=Supervised, Tuned on=RSICD2023.06 | 8.67 | 27.56 | 43.92 | 30.68 | — | |
| FGCLIPBackbone=ViT-B/162026.03 | 8.6 | 25.36 | 37.62 | — | — | |
| GeoRSCLIPBackbone=ViT-B/322026.03 | 8.45 | 25.01 | 38.35 | — | — | |
| GeoRSCLIPParadigm=GVLM + Pfeiffer, Tuned on=RS5M2023.06 | 8.12 | 24.81 | 38.79 | 23.94 | — | |
| FAAMIParadigm=Supervised, Tuned on=RSICD2023.06 | 8.11 | 25.59 | 41.37 | 23.18 | — | |
| SMLGN2026.07 | 7.85 | 27.14 | 42.58 | — | 149.21 | |
| KAMCL2026.07 | 7.78 | 25.23 | 40.02 | — | 150.52 | |
| GeoRSCLIPParadigm=GVLM + UniPELT, Tuned on=RS5M2023.06 | 7.52 | 23.35 | 35.66 | 22.39 | — | |
| LongCLIPBackbone=ViT-L/142026.03 | 7.14 | 23.48 | 35.55 | — | — | |
| SkyCLIPBackbone=ViT-B/322026.03 | 6.99 | 21.65 | 33.65 | — | — | |
| PIRParadigm=Supervised, Tuned on=RSICD2023.06 | 6.97 | 24.56 | 38.92 | 24.46 | — | |
| GeoRSCLIPParadigm=GVLM + LoRA, Tuned on=RS5M2023.06 | 6.73 | 23.24 | 35.59 | 21.33 | — | |
| LongCLIPBackbone=ViT-B/162026.03 | 6.68 | 20.68 | 32.17 | — | — | |
| SkyCLIPBackbone=ViT-L/14-pct302026.03 | 6.4 | 20.15 | 32.86 | — | — | |
| DOVEParms=38 M2023.10 | 6.04 | 23.95 | 40.35 | — | — | |
| SkyCLIPBackbone=ViT-L/14-pct502026.03 | 6.02 | 20.31 | 32 | — | — | |
| GeoRSCLIPParadigm=GVLM + Prefixtuning, Tuned on=RS5M2023.06 | 5.95 | 20.37 | 32.99 | 19.69 | — | |
| DOVE-SParms=25 M2023.10 | 5.84 | 21.87 | 36.54 | — | — | |
| CLIP-BaselineParadigm=GVLM, Tuned on=None2023.06 | 5.78 | 17.73 | 27.76 | 15.74 | — | |
| SWANParadigm=Supervised, Tuned on=RSICD2023.06 | 5.56 | 22.26 | 37.41 | 20.61 | — | |
| SWANParms=40 M2023.10 | 5.56 | 22.26 | 37.41 | — | — | |
| HVSAParadigm=Supervised, Tuned on=RSICD2023.06 | 5.51 | 21.13 | 34.13 | 20.16 | — | |
| HVSA2023.10 | 5.51 | 21.13 | 34.13 | — | — | |
| CLIPBackbone=ViT-B/322026.07 | 5.47 | 17.53 | 27.78 | — | 93.96 | |
| KCR2023.10 | 5.4 | 22.44 | 37.36 | — | — | |
| CLIPBackbone=ViT-B/162026.03 | 5.29 | 16.96 | 26.77 | — | — | |
| CLIPBackbone=ViT-L/142026.03 | 4.96 | 18.72 | 29.66 | — | — | |
| AFMFNParadigm=Supervised, Tuned on=RSICD2023.06 | 4.9 | 18.28 | 31.44 | 16.42 | — | |
| AMFMN2026.07 | 4.9 | 18.28 | 31.44 | — | 98.49 | |
| KCRParadigm=Supervised, Tuned on=RSICD2023.06 | 4.76 | 18.59 | 27.2 | 19.14 | — | |
| GaLRParadigm=Supervised, Tuned on=RSICD2023.06 | 4.69 | 19.48 | 32.13 | 18.96 | — | |
| GaLRParms=46 M2023.10 | 4.69 | 19.48 | 32.13 | — | — | |
| GaLR2026.07 | 4.69 | 19.48 | 32.13 | — | 113.78 | |
| LW-MCRParadigm=Supervised, Tuned on=RSICD2023.06 | 4.66 | 17.51 | 30.02 | 14.66 | — | |
| LW-MCR2023.10 | 4.66 | 17.51 | 30.02 | — | — | |
| VSE++Parms=29 M2023.10 | 4.37 | 15.37 | 25.35 | — | — | |
| CAMPParms=63 M2023.10 | 4.25 | 15.82 | 27.82 | — | — | |
| AMFMNParms=36 M2023.10 | 4.08 | 17 | 30.6 | — | — | |
| CAMERAParms=64 M2023.10 | 4 | 15.93 | 26.97 | — | — | |
| SCAN i2tParms=60 M2023.10 | 3.93 | 15.2 | 25.53 | — | — | |
| SCAN t2iParms=60 M2023.10 | 3.82 | 15.7 | 28.28 | — | — | |
| VSE++Paradigm=Supervised, Tuned on=RSICD2023.06 | 2.82 | 11.32 | 18.1 | 10.43 | — |